Fitness of Allozyme Variants

We have studied in Drosophila pseudoobscura the effect of allozyme variation on seven fitness components: female fecundity, egg hatchability, egg-toadult survival under near-optimal and under competitive conditions, rate of development under near-optimal and under competitive conditions, and mating capacity of males. Three genotypes at each of two loci, Pgm-1 and Me-2, have been studied in various combinations. These two loci are highly polymorphic in natural populations of D. pseudoobscura. Statistically significant differences involving one or more genotypes exist for all components of fitness. No single genotype is best for all fitness components; rather the relative fitnesses of genotypes are reversed when different parameters are considered, o r when they are studied in different environmental conditions. Also, the average egg-to-adult survival and rate of development are better when different genotypes are reared together than when they occur in pure culture. Four different modes of selection have been uncovered by our experiments. These forms of selection may account for the persistence of the two allozyme polymorphisms in nature, and for previously observed seasonal fluctuations of the allelic frequencies in natural populations. LECTROPHORETIC studies of genetic variation in natural populations of Edifferent species of Drosophila are now quite numerous (for a review of the literature up to 1972, see LEWOKTIN 1974) . A great deal of allelic variability has been found, and many authors have presented evidence indicating that much of the variation is maintained by balancing natural selection (e.g., PRAKASH and LEWONTIN 1968, 1971; AYALA, POWELL and TR~CEY 1972; AYALA et al. 1972, 1974; RICHMOND 1972; ZOUROS and KRIMBAS 1973; DOBZHANSKY and AYALA 1973). A number of laboratory studies have been directed toward ascertaining the mode of selection involved in maintaining allozyme polymorphisms. Changes in allelic frequencies due to natural selection have been demonstrated at the Mdh-2 locus in D. willistoni, D. equinoxialis, and D. tropicalis (AYALA and ANDERSON 1973). In D. willistoni there is also evidence of selection at the Lap-5 and Est-5 loci, but not at the sex-linked, Est-7 (POWELL 1973). YAMAZAKI (1971) 'Work suppoited by U S Reprint requests fiom Europe should be addressed to D MARINKOVIC, from elsewhere, to F J AYALA Public Health Service Research Fellowshlp (1 F05 TWO 1991-01) t o D M , and by Grant AT (04-3) 34 from the U S Atomic Energy Commission to F J A Genehcs 79: 8:-95 Januaij , 1975 86 D. M A R I N K O V I ~ AND F. J. AYALA found no evidence, either, of natural selection at the sex-linked esterase locus, Est-5. in D. pseudoobscura, although FONTDEVILA (1973 and personal communication) has recently found evidence of selection at the locus. The difference between these two reports may be due to the fact that YAMAZAKI’S study was started with strains of flies kept in the laboratory for many years before the beginning of the experiment, while FONTDEVILA’S experimental populations were started with the descendants of flies recently collected in nature. MACINTYRE and WRIGHT (1966) obtained inconclusive results at a sex-linked esterase locus, Est-6, in both D. melanogaster and D. simulans. RASMUSON, RASMUSON and NILSON (1967) studied experimental populatiom of D. melanogaster with different initial frequencies of the Est-6” and Est-@ alleles; the authors considered it probable that the Est-6 polymorphism was maintained by balancing selection affected by the genetic background. In D. virilis, OHBA (1969) found no significant frequency changes at two loci, Est-2 and Est-7, although changes occurred at the Est-9 locus. There are as yet very few studies attempting to asc-xtain what fitness components are affected by the electrophoretic variants. KOJIMA and YARBROUGH (1967) and KOJIMA and TOBARI (1969) have reported that genotypic fitnesses for viability are frequency-dependent at the Est-6 and Adh loci in D. melanogaster. On the other hand, YAMAZAKI (1971) found no significant differences ir? several fitness components among Est-5 genotypes of D. pseudoobscura. The experiments reported here were made to ascertain whether fitness components in D. pseudoobscura are affected by genotypic variation at two loci, PgmI and Me-2. DOBZHANSKY and AYALA (1973) have shown that allelic frequencies at these two loci fluctuate through the seasons in two natural populatior?s, and that the fluctuations are repeated from one year to the next. The oscillations observed by DOBZHANSKY and AYALA might conceivably be due to differential selective values at different stages of the life cycle, perhaps associated with changes in population density o r with environmental parameters. At each of the two loci studied by DOBZHANSKY and AYALA, only two alleles occur in substantial frequencies, and together have a frequency of about 95 % or higher. Our studies are concerned with these two common alleles at the Pgm-1 and Me-2 loci. MATERIALS A N D METHODS The Pgm-1 locus is sex-linked while Me-2 is in the second chromosome. These loci code for the enzymes, phosphoglucomutase-l and malic enzyme-2, respectively. Our techniques of starch gel electrophoresis and the enzyme assays used have been described elsewhere (AYALA, POWELL and TRACY 1972). WP have used several strains descended from flies collected in the fall of 1972 at MacDonald Ranch, Napa County, California. Several thousand inseminated wild females were placed in individual cultures. Single-pair sib matings were made with the F, progenies to establish the experimental strains. The genotypes of the two parents were ascertained by electrophoresis, and strains having the desired genetic ccnstitutions were chosen. In some cases, a second generation of sib pair matings was obtained. There are three groups of strains. The first one (group A) consists of 10 strains, each homozygous for Pgm-1’04 and for Me-2100. The second one (B) consists of 6 strains homozygous for Pgm-1100 and Me-2102. The third group (C) consists of 12 strains with flies homozygous for Pgm-1100 and Ms-2’00. The number of strains simultaneously FITNESS OF ALLOZYMES IN D. pseudoobscura 87 homozygous for the two rare alleles (Pgm-1104 and Me-2102) was too small, and we decided not to use them. Each one of our strains descends from a different wild inseminated female. At the beginning of our experiment, the flies were fourth or fifth generation descendants of the wild flies. To randomize the genotypes, series of crosses between the strains of each group have been performed over two generations. If the strains of group A are represented as A,, A,, A,, etc., we have first crossed A, x A,, A, x A,, etc.; then, we have crossed Al..2 x A,_,, A2-, X A,-,, etc. Thus, we have obtained within each group five combined strains, each originating from four different ancestral strains. In the third generation, crosses have been made between pairs of such synthetic strains within the same group (e.g., A,-* x As-*; or B,-6 x Bl-2; or C,-, X C,-,, etc.), and also between strains from different groups (e.g., AI+ x B1--4; B,-s X C2-,; or x C,-,, etc.). Consequently, six different groups, each with five synthetic strains, have been obtained. Each of these synthetic strains is derived from at least six (B), and generally eight, original strains. All synthetic strains should have much of the genetic variation in the original population, except for the two loci under consideration. The genotypes of the synthetic strains at the Pgm-l and Me-2 loci are as follows: A B C A x B A x C B x C Pgm-I 104/104 100/100 100/100 104/100 104/100 100/100 Me-2 100/100 102/102 100/100 100/102 100/100 102/100 The 30 synthetic strains (five in each of the six groups) have been raised simultaneously, and hatchability of their eggs, egg-to-adult survival, and rate of development have been measured under near-optimal conditions (i.e., 100 eggs per half-pint bottle, with two replications per strain). Also, egg-to-adult survival and rate of development have been measured under competitive conditions (i.e., 200 eggs per 11 mm-diameter vial, with one replica per strain). The females raised under optimal conditions have been compared for fecundity, by measuring the number of eggs laid by each of two groups of 3 females with 3 males from each strain, from the 6th to the 13th day after emergence (see MARINKOVIC 1967). Males, raised under optimal conditions, were compared for their mating ability; males from each strain were placed individually in a bottle with 8 virgin females taken from a common pool oNf all 30 synthetic strains. Number of females inseminated by each male after 24 hours was determined by inspecting the reproductive tracts of the females for presence of sperm. Egg-to-adult survival and rate of development have also been measured in two sets of vials, each vial containing 198 eggs, In the first se& the eggs in a vial all came from the five synthetic strainq of a given group. In the second set, 33 eggs came from each of the six groups of strains. All cultures were kept in a constant temperature incubator at 25 * 0.5" and ca. 70% relative humidity. The food was standard agar-cornmeal-molasses Drosophila medium. The following should be noted concerning the statistical methods. We are studying three genotypes at each of two loci; nine different genotypes are possible when both loci are considered simultaneously. Owing to the difficulty of obtaining enough strains homozygous simultaneously for the two rare alleles, we have used in our experiment only six of the nine possible genotypes. Therefore, it is not possible to carry out a straightforward two-way analysis of variance to separate the effects of the two loci and to detect interactions between them. Instead, we have performed a series of nonortho'gonal linear contrasts for each of the two most common genostypes at each locus. and for the interaction between the two loci. This procedure was advised by a reviewer of a previous version of this paper. For each fitness component we have tested five hypotheses, namely whether each of P g m l l o o / l o o , Pgm-1100/104, Me-2100/100, Me-21Oo/'02, and the interaction, cause a significant deviation from the mean effect of all the genotypes. The following matrix of coefficients was used for the nonorthogonal comparisons. (See above for the genotypes of each group of strains.) 88 D. MARINKOVI~: AND F. J. AYALA Genotype Strains tested A B C A x B A x C B x C p g m . ~ ' O O / ' O O 4 0 7 -5 -3 5 pgm.110n/1n4 -4 0 -5 7 9 -7 ~ ~ 2 1 O O / ' O O 0 -4 7 -5 5 -3 Me.2100/10% 0 -4 -5 7 -7 9 Interaction 0 0 -6 -6 6 6 The degrees of freedom for each contrast are 1, 6(n-1), where n is the number of replications made in the study of a given fitness component. In the few cases where a replicate was missing for a given genotype, we have replaced it by the mean value of the other replicates, and reduced accordingly the degrees oE freedom of the denominator.